THIN REFRACTORY METAL LAYER USED AS CONTACT BARRIER TO IMPROVE THE PERFORMANCE OF THIN-FILM SOLAR CELLS

Abstract

A thin film amorphous silicon solar cell may have front contact between a hydrogenated amorphous silicon layer and a transparent conductive oxide layer. The cell may include a layer of a refractory metal, chosen among the group composed of molybdenum, tungsten, tantalum and titanium, of thickness adapted to ensure a light transmittance of at least 80%, interposed therebetween, before growing by PECVD a hydrogenated amorphous silicon p-i-n light absorption layer over it. A refractory metal layer of just about 1 nm thickness may effectively shield the oxide from the reactive plasma, thereby preventing a diffused defect when forming the p.i.n. layer that would favor recombination of light-generated charge carriers.

Description

FIELD OF THE INVENTION

This disclosure relates to photovoltaic solar cells and, in particular, to thin film amorphous silicon solar panels and related processes.

BACKGROUND OF THE INVENTION

Thin-film solar cells based on hydrogenated amorphous silicon (a-Si:H) are a promising photovoltaic technology for delivering low-cost solar energy. They can be used for cost-effective applications, such as large area photovoltaic modules and cells on flexible substrates as well. The silicon is deposited at low temperatures (<200° C.) by using plasma-enhanced chemical vapor deposition (PECVD). This deposition technique is effective and enables the deposition on large area substrates with good uniformity. Moreover, deposition conducted at very low temperatures enables the utilization of many different types of substrates, such as metals, glass, plastics etc. Unfortunately, the use of plasma inevitably affects the substrate material, often causing a diffused defect at the transparent front contact that increases recombination of light-generated charge carriers.

SUMMARY OF THE INVENTION

An approach is for improving the above mentioned drawback for a plasma-enhanced chemical vapor deposition fabrication technique of hydrogenated amorphous silicon (a-Si:H) solar panel.

Typically, transparent conductive oxide layers (TCO), such as ZnO or SnO₂ films, are used as front and in some cases also as back contacts for the p-i-n cells. In these devices, the interface between contact layers and a-Si:H plays an important role on cell performances. See F. Smole et al. J. Non Cryst. Solids 194, 312 (1996); Vinh Ai Dao et al., Solar Energy 84, 777 (2010); and J. S. C. Prentice et al. J. Non-Cryst. Solids, 262, 99 (2000). For better insight of this, the behavior of solar cell structure based on the stack sequence SnO₂:F/a-Si:H/Mo, which is typically used in commercial thin film silicon based solar cells, has been studied. The total capacitance of the studied structure may be modeled as a series of depletion capacitances of the two junctions connected back-to-back, i.e. at the transparent front contact, the junction SnO₂:F/a-Si:H and, at the back contact, the Mo/a-Si:H junction.

These verifications and analysis were prompted by a rewarding trade-off that could be reached between the ability of a refractory metal to withstand plasma aggression and possibly shield a more delicate TCO surface layer of a substrate of deposition of the hydrogenated amorphous silicon thin film, and the transparency of such refractory metal barrier to the light. Besides proving a remarkable effectiveness of the original intuition, it was found that even the contact band-offset of such an ultra-thin-metal/a-Si:H semiconductor interface that also could have a non-negligible effect on overall energy conversion efficiency at a modified interface.

Basically, a very thin layer of a strongly refractory metal, such as molybdenum, tungsten, tantalum or titanium, as an interlayer between a TCO contact layer and the a-Si:H layer of a p-i-n structure of a solar cell, may improve its performance. The structure disclosed herein may apply to and may be effective even in a so-called superstrate configuration of a common metal-back contact structure of a solar panel. A refractory metal is robust against the plasma action during the a-Si:H deposition; however, to be effective, the layer must be thinner than the “skin depth” of the metal, but sufficiently thick to act as a barrier to the plasma ions that are typically accelerated with a potential of about 100V.

The presence of a thin metal layer at the contact interface triggers the excitation of surface plasma polaritons (SPPs), the effects of which in improving light transmission and light capture in the absorption silicon layer of the cell will be discussed later in this description. For example, the refractory metal may comprise molybdenum, because it has adequate robustness to plasma aggression and light transmission higher than 85% at optimal thickness. Additionally, tungsten has also been found to be a viable alternative candidate, suggesting that even other refractory metals like tantalum and titanium may be employed.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1a shows simulated and experimental C-V curves related to a double junction structure SnO₂:F/a-Si:H/Mo and a schematic diagram of the double junction structure, according to the prior art.

FIG. 1b shows the estimated energy band diagram of the double junction structure of FIG. 1a.

FIG. 2 are J-V characteristics recorded under illumination and dark conditions for a SnO₂:F/p-type-a-Si:H junction and for a Mo/p-type-a-Si:H junction, respectively, according to the prior art.

FIG. 3a shows a superstrate configuration of a hydrogenated amorphous silicon (a-Si:H) solar cell, according to the prior art.

FIG. 3b shows a superstrate hydrogenated amorphous silicon (a-Si:H) solar cell structure with a thin interlayer of refractory metal between the TCO layer and the p-i-n layer of amorphous silicon, according to the present invention.

FIG. 4 shows Re(∈) and Im(∈) characteristics of molybdenum as measured with REELS (circles) and calculated using the Drude model (lines), according to the present invention.

FIG. 5 shows diagrams of optical constants of molybdenum vs. wavelength, experimental values of n and k as a function of the wavelength and energy for molybdenum, tantalum and tungsten, according to the present invention.

FIG. 6 shows experimental values of n and k as a function of the wavelength and energy for molybdenum, tantalum and tungsten, according to the present invention.

FIG. 7 shows dispersion curves (energy vs. wave-vector k_SPP): from the Drude's model, at the top, and when effective values of Re(∈) and Im(∈) are considered, at the bottom, according to the present invention.

FIG. 8 shows diagrams of lateral skin depth of SPP modes as a function of the wavelength evaluated by starting from experimental values of ∈: the ideal case of the Drude's theory at the top and the experimental case at the bottom, according to the present invention.

FIG. 9 shows diagrams of skin depth vs. wavelength for both SPP and Plasmon modes and for the evanescent waves in the metal: calculated according the Drude's model, at the top, and obtained starting from the experimental n and k values, at the bottom, according to the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

The experimental C-V and I-V characteristics of FIG. 1a and of FIG. 2, respectively, were both modeled by assuming a high interface state density and using the built-in voltages (V_bi) of the two junctions as fitting parameters. FIG. 1a shows the simulated and experimental C-V curves related to a typical structure comprising the double junction structure SnO₂:F/a-Si:H/Mo (in the example shown, the a-Si:H is a p-type layer), as sketched in the inset. The simulated C-V curves are for three different cases depending on the relative values of V_bi,1 and V_bi,2. If V_bi,1>V_bi,2, the capacitance decreases; if V_bi,1<V_bi,2, the capacitance increases. If V_bi,1=V_bi,2, the capacitance is almost constant. The experimental data is in agreement with simulated curves assuming V_bi,SnO2:F=0.16V and V_bi,Mo=0.14V.

FIG. 1b shows the estimated energy band diagram of the junction Mo/p-type-a-Si:H and of the junction SnO₂:F/p-type-a-Si:H. In order to explain the experimental I-V characteristics, a Gaussian surface state density of 4.0×10¹³ cm⁻², located 0.4 eV above the valence band edge of the p-type a-Si:H layer, must be assumed. See G. Cannella et al. in press on J. Appl. Phys. (2011). The V_bi of the SnO₂:F/p-type-a-Si:H interface is higher than that of the Mo/p-type-a-Si:H interface, with a difference of about 20 mV. This means that molybdenum provides a better Schottky contact with p-type-a-Si:H than SnO₂:F.

Comparative test samples of thin film a-Si:H solar cells with p-i-n configuration were fabricated, some using a SnO2:F and other molybdenum rear contacts with the p-type-a-Si:H film. In both types of test samples, the front contact of the cells was identical and included a ZnO:Al transparent conducting layer.

Analysis was done by measuring cell parameters at AM1.5G and in dark. The utilized figures were averaged over 100 devices of each type. With reference to FIG. 4, where a comparison of typical J-V curves for the two cell types is illustrated, an improvement of about 100 mV in the open circuit voltage (V_oc) can be observed for the cell with molybdenum contact compared to the cell with SnO₂:F contact. Such an improvement cannot be explained only in terms of barrier height difference (˜20 meV). More insight is provided by dark J-V characteristics shown in FIG. 2. Indeed, a much lower diode saturation is observed in the molybdenum contact samples (one decade lower than in the SnO₂ contact samples), which is consistent with the improvement of V_oc. It may be suggested that in addition to the lower barrier height, a lower carrier recombination takes place in the case of a Mo/p-type a-Si:H junction, likely due to a lower defect density in the intrinsic a-Si:H layer.

Molybdenum is a highly refractory material and suffers less from plasma damaging during the PECVD growth of a-Si:H p-i-n layer on it. This allows reduction in the defects that are generated when the contact is exposed to the high reactive plasma environment, necessary for the deposition of the p-i-n layer. Therefore, compared with the more vulnerable non-SnO₂:F contact, when using molybdenum as a substrate for PECVD, a lesser number of the plasma-generated impurities migrate in the sensing regions of the p-i-n layer (particularly in the intrinsic layer), thus promoting a longer life-time and lower recombination of photon-generated electron-hole pairs.

This is particularly effective in the substrate configuration of thin film solar cells where the front electrode (the electrode exposed to the light), which typically is a transparent conductive layer (SnO₂, ZnO, ITO, etc.), is deposited at the end of the fabrication process, after the reactive plasma process (PECVD) of deposition of the p-i-n a-Si:H layer. Differently, for a so-called superstrate configuration, in which the front electrode TCO is the substrate of deposition of the p-i-n a-Si:H layer, the situation changes. In fact, TCO is not a refractory material and is subject to aggression by the reactive plasma.

A typical superstrate configuration 20 is depicted in FIG. 3a. The superstrate configuration 20 illustratively includes a glass layer 26, a TCO layer 25 thereon, p-type silicon 24 on the TCO layer, an intrinsic zone 23 thereon, n-type silicon 22 on the intrinsic zone, and a top contact 21 on the n-type silicon. In a conventional p-i-n superstrate structure, the light illuminates the cell from the bottom. A textured TCO layer 25 is generally used as contact in order to enhance a light trapping effect. TCO is typically made of SnO₂:F or ZnO:B or ZnO:Al, with a resistivity of 0.8-1×10⁻³ Ohm cm and an adequate transmittance in the wavelength range between 350 nm and 1,200 nm.

The superstrate structure 30 of the present disclosure is depicted in FIG. 3b. The superstrate configuration 30 illustratively includes a glass layer 36, a TCO layer 35 thereon, an interlayer refractory metal 34 on the TCO layer, p-type silicon 33 on the refractory metal, an intrinsic zone 37 thereon, n-type silicon 32 on the intrinsic zone, and a top contact 31 on the n-type silicon. The structure has an interlayer 33 of a refractory metal that may be chosen among molybdenum, tungsten, tantalum and titanium, of thickness adapted to ensure adequate transmittance, between the TCO and the p-i-n layer. The process of the present disclosure, compared to a normal flow of the typical fabrication process of a superstrate structure, includes an additional step of depositing on top of the TCO (front contact layer) a very thin layer of refractory metal, molybdenum in the shown example, that has the function of reinforcing the conductive contact layer on which the p-i-n layer, typically of hydrogenated amorphous silicon, is successively grown.

These findings lead to significant enhancement of the contact barrier, making it more resistant against the reactive plasma used for successively depositing/growing the p-i-n layer. Improvement of the contact leads to a better contact/a-Si:H barrier (improved band-offset) as well as decreasing recombination rate because of a reduced density of defects in the semiconductor phonon absorption region.

In order to ensure an adequate transmission of the light, it is important to limit the thickness of the refractory metal interlayer. Thereby, it is necessary to establish the best trade-off between positive effects on V_oc and transmittance of the interlayer.

Physics Background for the Exemplary Case of a Molybdenum Interlayer

FIG. 4 shows the real and imaginary parts of the dielectric constant of Molybdenum, measured by Reflection Electron Energy Loss Spectroscopy—(REELS) (circles) and calculated by using the classical Drude model. See W. A. Harrison, “Solid State Theory,” Dover Publication Inc, New York, 1979. The dielectric constant of metals in the visible range is negative because of the response of conduction electrons, as explained by the Drude model (lines in the figure). The dielectric constant is then zero in correspondence of the plasma frequency and becomes positive at high frequencies tending to one in the UV range.

The Drude model explains the real part of the dielectric constant; however, the imaginary part deviates remarkably from the experimental values (FIG. 4). In fact, the experimental value is much higher than that expected from the Drude theory. In other words, the dissipative part of the radiation propagating in the metal is much higher.

From the dielectric constant ∈, the complex refracting index n* can be derived. The refractive index can be written as:

n*=n+ik;  (1)

where n and k are the real and imaginary part of the refractive index.
The absorption coefficient can be written as:

$\begin{array}{cc}\alpha =\frac{4\pi k}{\lambda }=\frac{1}{\delta };& \left(2\right)\end{array}$

where λ is the light wavelength in vacuum, and δ is the skin depth of the metal (molybdenum).

FIG. 5 shows the experimental values of Re(∈), Im(∈), n and k as a function of the wavelength in vacuum. From the optical constants, the metal response to the radiation can be evaluated. When the frequency is high, i.e. greater than the plasma frequency (υp) value, the dielectric constant of the metal is positive and the radiation can propagate in the metal, giving rise to longitudinal wave modes that propagate laterally along the metal-dielectric interface. Such modes, defined as surface plasma polaritons couple optical radiation and plasma electrons collective oscillations known as plasmons. See C. Kittel, “Introduction to Solid State Physics,” Wiley (1986), p. 253.

To very high frequencies, the metal becomes transparent (n→1 and k→0). On the other hand, when the dielectric constant is negative and assumes high values (for frequencies υ<υp), the reflectivity is very large and the transmitted component of the radiation originates evanescent waves, which propagate in depth for very small distances.

FIG. 6 shows a comparison of molybdenum with the other refractory metal candidates, such as Ta and W, which exhibit a behavior quite similar to the case of molybdenum. At the vacuum/metal interface, or more in general at the dielectric/metal interface, longitudinal wave modes are generated. These collective oscillations propagate laterally along the metal/dielectric interface. Such longitudinal modes couple the optical radiation with oscillating collective modes of conducting electrons of the metal, which are defined as surface plasmon-polaritons (spp) modes.

According to Drude's theory, the dispersion relation which relates the wave-vector

$\begin{array}{cc}{k}_{\mathrm{spp}}=\frac{2\pi }{{\lambda }_{\mathrm{spp}}}& \left(3\right)\end{array}$

of the SPP modes to the frequency ω (where ω=2πυ), is shown at the top of FIG. 7.

In the figure, the top part corresponds to the plasmon part and the bottom part is that of polaritons, SPPs. When the effective values of Re(∈) and Im(∈) are considered, the dispersion curve changes significantly, and the behavior is shown at the bottom of the figure. Plasmon and SPP modes have much smaller extinction lengths, i.e. they are quenched very quickly, propagating in the metal (plasmons) or along the dielectric/metal interface (polaritons).

In particular, at the bottom of FIG. 8, the lateral extinction length of SPP modes as a function of the wavelength in vacuum evaluated by using the experimental values of the dielectric constant is shown. The ideal behavior calculated from Drude's model is shown at the top of the figure, where it can be observed that the ideal value of extinction length is much higher.

In a similar way as in FIG. 8, FIG. 9 depicts the skin depths δ as a function of the wavelength (in vacuum) both for SPP and plasmon modes and for the evanescent waves in the metal. For SPP modes, the figure shows the extinction lengths both in the dielectric and in the metal. Because SPPs are surface modes, they are localized only at the interface, whereas they vanish in the perpendicular direction, along the direction of the dielectric and of the metal. Note that the extinction depth of SPP is higher in the dielectric than in the metal over the whole spectrum, starting from 200 nm. This suggests that the radiation is confined mainly in the dielectric part, with a significant light trapping effect. The values of the skin lengths in the metal in a range of wavelengths between 200 nm and 1 μm are about 6 nm, for evanescent waves, and 30-40 nm for SPPs.

In the structure depicted in FIG. 3b, when the light impinges on the interface TCO/Mo, it will give rise to evanescent waves and to SPP modes that propagate laterally along the interface. If the Mo layer is thin enough, the attenuation produced by the metal layer is sufficiently low, leading to an acceptable transmission of the radiation. It is possible to estimate an adequate thickness for the Mo layer, which is of about 1 nm. Considering only the effect of the evanescent waves, the attenuation would be at most of about exp(−1 nm/6 nm), which corresponds to about 85%.

Actually, it is expected that the value may be higher than 85% due to the effects of the forming of SPPs that propagate laterally along the Mo/a-Si:H interface. Reasonably, it is expected that if the TCO surface is textured, as it is a typical practice in the art, hence with a relatively high roughness, even if the radiation hits the surface orthogonally (as in the case of solar cells under optimal orientation conditions), it will promote the excitation of both evanescent optical radiation waves (evanescent waves) and SPP modes (surface plasma polaritons). Thereby, the intensity of radiation passing through the Mo layer and reaching the semiconductor will be higher than the contribution due to the evanescent waves only. That is, the effective transmittance of a molybdenum layer of about 1 nm thickness is expected to be higher than 85% (which is the contribution if only the evanescent waves are considered). On the other hand, beneficial effects by a Mo layer of 1 nm on the Voc are expected. Indeed, from an electrostatic viewpoint, a thin metal layer of 1 nm is equivalent to a bulk metal, considering that the Thomas-Fermi screening length in a metal like molybdenum is of 0.05 nm.

As far as the screening function of the refractory metal interlayer from the reactive plasma of the PECVD process of the p-i-n layer is concerned, considering that the “projected range” and the “longitudinal straggling” of Si ions of about 100 mV in molybdenum are of about 0.5 nm and of about 0.6 nm, respectively, it may be reasonably concluded that a molybdenum layer of 1 nm is able to effectively shield the underlying TCO from the reactive plasma and thus prevent formation of impurities during the PECVD growth of the p-i-n silicon layer.

Basically, the fabrication process of thin film amorphous silicon solar cells comprising at least one of front and rear cell contacts between a hydrogenated amorphous silicon layer and a transparent conductive oxide layer, comprises the sequence of steps of: depositing a transparent conductive oxide layer for either a front or a rear cell contact; depositing a layer of refractory metal, chosen among the group including molybdenum, tungsten, tantalum and titanium, of thickness adapted to ensure a light transmittance above 80%, over a contact surface of the transparent conductive oxide layer; and depositing by plasma enhanced chemical vapor deposition hydrogenated amorphous silicon as far as growing a cell p-i-n layer, over the refractory metal contact interlayer.

By texturing of the surface of the deposited layer of transparent conductive oxide, typically sub-oxides of Zn or of Sn (i.e. non stoichiometric ZnO or SnO₂) may be produced for enhancing incident light trapping according to a common practice in the art. Texturing is commonly done by etching the surface of the deposited layer of transparent conductive oxide with HCl. Preferably, a desirable morphology of the textured surface is pre-tuned by varying parameters like pressure, temperature and gas flow rate in the deposition chamber during deposition of the oxide. Sputtering, chemical vapor deposition techniques and sequential chemical vapor deposition of atom thick layers techniques may be used for depositing the refractory metal layer over the layer of transparent conductive oxide.

A highly conformal layer of refractory metal may be obtained even on a textured surface of transparent conductive oxide with low pressure chemical vapor deposition (LP-CVD) or even more so by sequential chemical vapor deposition of atom thick layers as, for example, with the so called atomic layer deposition (ALD) technique. For background, see N. Ashcroft, N. D. Mermin, “Solid State Physics”, Saunders College Publishing, Fortworth, Philadelphia, (1976) p. 551.

Claims

1-12. (canceled)

13. A method for making a thin film amorphous silicon solar cell comprising a hydrogenated amorphous silicon layer, and a transparent conductive oxide layer under the hydrogenated amorphous silicon layer, the method comprising:

depositing the transparent conductive oxide layer for a cell contact;

depositing a layer of refractory metal comprising at least one of molybdenum, tungsten, tantalum, and titanium, with a thickness for a light transmittance of at least 80%, the layer of refractory metal being over a contact surface of the transparent conductive oxide layer; and

plasma enhanced chemical vapor depositing hydrogenated amorphous silicon to grow a cell p-i-n layer, the hydrogenated amorphous silicon being over the layer of refractory metal.

14. The method of claim 13 wherein the layer of refractory metal comprises molybdenum; and wherein depositing the layer of refractory metal comprises depositing the layer of refractory metal to form a continuous layer having a thickness of about 1 nm.

15. The method of claim 13 wherein depositing the layer of refractory metal comprises sputtering the layer of refractory metal.

16. The method of claim 13 wherein depositing the layer of refractory metal comprises chemical vapor depositing the layer of refractory metal.

17. The method of claim 13 further comprising texturing a surface of the transparent conductive oxide layer before depositing the layer of refractory metal.

18. The method of claim 17 wherein texturing comprises hydrochloric acid etching the transparent conductive oxide layer.

19. The method of claim 17 wherein texturing morphology is adjusted during deposition by chemical vapor deposition of the transparent conductive oxide layer by selectively setting pressure, temperature, and gas flow rate in a deposition chamber.

20. The method of claim 13 wherein depositing the layer of refractory metal comprises low pressure chemical vapor depositing.

21. The method of claim 13 wherein depositing the layer of refractory metal comprises sequential chemical vapor depositing of atom-thick layers of the refractory metal.

22. A method for making a solar cell comprising an amorphous silicon layer, and a transparent conductive oxide layer under the amorphous silicon layer, the method comprising:

forming the transparent conductive oxide layer for a cell contact;

forming a layer of refractory metal over a contact surface of the transparent conductive oxide layer; and

forming amorphous silicon over the layer of refractory metal.

23. The method of claim 22 wherein depositing comprises plasma enhanced chemical vapor depositing the hydrogenated amorphous silicon to grow a cell p-i-n layer.

24. The method of claim 22 wherein the layer of refractory metal comprises at least one of molybdenum, tungsten, tantalum, and titanium.

25. The method of claim 22 wherein the layer of refractory metal has a thickness for a light transmittance of at least 80%.

26. The method of claim 22 wherein the layer of refractory metal comprises molybdenum; and wherein forming the layer of refractory metal comprises depositing the layer of refractory metal to form a continuous layer having a thickness of about 1 nm.

27. The method of claim 22 wherein forming the layer of refractory metal comprises sputtering the layer of refractory metal.

28. The method of claim 22 wherein forming the layer of refractory metal comprises chemical vapor depositing the layer of refractory metal.

29. The method of claim 22 wherein forming the layer of refractory metal comprises low pressure chemical vapor depositing.

30. The method of claim 22 wherein forming the layer of refractory metal comprises sequential chemical vapor depositing of atom-thick layers of the refractory metal.

31. A method for making a silicon solar cell comprising an amorphous silicon layer, and a transparent conductive oxide layer under the amorphous silicon layer, the method comprising:

forming the transparent conductive oxide layer for a cell contact;

forming a layer of refractory metal over a contact surface of the transparent conductive oxide layer to have a light transmittance of at least 80%;

forming amorphous silicon over the layer of refractory metal; and

texturing a surface of the transparent conductive oxide layer before depositing the layer of refractory metal.

32. The method of claim 31 wherein depositing comprises plasma enhanced chemical vapor depositing the hydrogenated amorphous silicon to grow a cell p-i-n layer.

33. The method of claim 31 wherein the layer of refractory metal comprises at least one of molybdenum, tungsten, tantalum, and titanium.